Available online at www.sciencedirect.com

ch 62 (2009) 61–67

Journal of Business Resear

Information direction, website reputation and eWOM effect:A moderating role of product type

Cheol Park a,1, Thae Min Lee b,⁎

a School of Business Administration, Korea University, Jochiwon-Eup, Yongi-Gun, Chungnam, 339-700, South Koreab Department of Business Administration, Chungbuk National University, Gaeshin-Dong, Heungduk-Gu, Cheongju, Chungbuk, 361-763, South Korea

Received 1 December 2006; received in revised form 1 October 2007; accepted 1 November 2007

Abstract

This article examines how the electronic word of mouth (eWOM) information direction (positive vs. negative) and a website's reputation(established vs. unestablished) contribute to the eWOM effect. The article describes a study focusing on the moderating role of the product type(search vs. experience). The results of the experiment show that the eWOM effect is greater for negative eWOM than for positive eWOM, greaterfor established websites than for unestablished websites, and greater for experience goods than for search goods. The results support themoderating effects of product type on the eWOM information direction-website reputation-eWOM effect relationship. The impact of negativeeWOM on the eWOM effect is greater for experience goods than for search goods. Similarly, the impact of website reputation on the eWOM effectis greater for experience goods than for search goods. The findings provide managerial implications for an Internet marketing strategy.© 2008 Elsevier Inc. All rights reserved.

Keywords: Internet marketing; Electronic word of mouth (eWOM); Direction; Website reputation; Search goods; Experience goods

1. Introduction

The arrival and expansion of the Internet has extendedconsumers' options for gathering product information byincluding other consumers' comments, posted on the Internet,and has provided consumers opportunities to offer their ownconsumption-related advice by engaging in electronic word-of-mouth (eWOM) (Hennig-Thurau et al., 2004). Gruen, Osmon-bekov and Czaplewski (2006) suggest that the customer-to-customer know-how exchange, a specific form of eWOM, has a

⁎ Corresponding author. Lee is to be contacted at Department of BusinessAdministration, Chungbuk National University, Gaeshin-Dong, Heungduk-Gu,Cheongju, Chungbuk, 361-763, South Korea. Tel.: +82 43 261 2334; fax: +8243 276 2347.

E-mail addresses: cpark@korea.ac.kr (C. Park), leetm@chungbuk.ac.kr(T.M. Lee).1 Park, School of Business Administration, Korea University, Jochiwon-Eup,

Yongi-Gun, Chungnam, 339-700, South Korea. Tel.: +82 41 860 1566; fax: +8241 860 1507.

0148-2963/$ - see front matter © 2008 Elsevier Inc. All rights reserved.doi:10.1016/j.jbusres.2007.11.017

direct relationship with loyalty intentions, as well as an indirectrelationship mediated through the overall value of the firm'soffering. Previous research on traditional (offline) WOMsuggests that factors such as source credibility (Dholakia andSternthal, 1977) and WOM directions (positive or negative)(Arndt, 1967) are antecedents of the WOM effect.

In eWOM however, unlike the case of WOM frominterpersonal sources, recommendations are typically fromunknown individuals and in a text-based format. So, onlineconsumers have difficulty in using source similarity todetermine the credibility of information (Chatterjee, 2001).Therefore, in order to build effective Internet marketingstrategy, marketers should understand the eWOM effect, notonly in terms of source credibility, but also in terms of themessage configuration (negative/positive eWOM posted on theestablished/unestablished website). However, little researchaddresses how eWOM message configurations affect theeWOM effect. Therefore, the vehicle effect is considered byincorporating a website's reputation (Shamdasani et al., 2001)as a determinant of the credibility of the eWOM-specific source.

62 C. Park, T.M. Lee / Journal of Business Research 62 (2009) 61–67

In addition, this study suggests the eWOM informationdirection as a message characteristic variable.

One interesting, but unanswered question, is does the impactof the eWOM information direction and website reputation onthe eWOM effect vary with the product type? In order toaddress this void, this study focuses on the moderating effect ofproduct type (search vs. experience) in investigating the impactof eWOM information direction and website reputation on theeWOM effect.

Therefore, the objectives of this study include the followingpoints. First, to examine the impact of the eWOM informationdirection (positive vs. negative eWOM) on the eWOM effect;second, to investigate the influence of website reputation(established vs. unestablished website) on the eWOM effect;and finally, to examine how the interactions of eWOMinformation direction and website reputation with producttype (search goods vs. experience goods) influence the eWOMeffect.

2. Conceptual background and research hypotheses

2.1. eWOM direction

A significant finding in impression formation literature is thenegativity effect, namely that people place more weight onnegative information rather than positive information in thesubjects' forming overall evaluations (Skowronski and Carl-ston, 1989). Evidence suggests that negative information ismore attention grabbing in general and receives greater scrutinythan positive information (Homer and Yoon, 1992). Previousresearch suggests that the WOM influence appears to beasymmetrical in that a negative WOM has a stronger influenceon customers' brand evaluations (Arndt, 1967; Mizerski, 1982;Richins, 1983; Wright, 1974) and on the purchase intentions ofpotential buyers (Brown and Reingen, 1987; Weinberger et al.,1981) than a positive WOM. These arguments offer a rationalefor the following hypothesis. H1: The eWOM effect is greaterfor negative eWOM information than for positive eWOMinformation.

2.2. Website reputation

Source variables, whether in the form of a messagespokesperson or a media vehicle itself essentially moderatethe impact of a persuasive message by influencing theaudience's confidence based on the endorsement by the source(Shamdasani et al., 2001). This vehicle source effect is likely tobe applied to the eWOM context. Consumer confidence broughtabout by a credible website might influence the eWOM postedon the website.

According to Shamdasani et al. (2001), due to low barriers ofentry and lack of gatekeepers on the Web, well-established andreputable websites have been more readily accepted byconsumers than have unknown sites. The previous discussionsuggests the following hypothesis. H2: The eWOM effect isgreater for websites with established reputations than for thosewithout established reputations.

2.3. Categorization of goods: search vs. experience

According to Nelson (1974), search goods are defined asthose characterized by product attributes where completeinformation about the goods can be acquired prior to purchase;experience goods are characterized by attributes that cannot beknown until the purchase and after use of the product or forwhich an information search is more costly and/or difficult thandirect product experience (Klein, 1998). Maute and Forrester(1991) examine the applicability of this search/experienceframework to the information search process. Maute andForrester (1991) suggest search and experience qualities asmoderators of the link between search antecedents and externalsearch efforts. By extending this framework to the study of theinfluence of eWOM information content on the eWOM effectacross product categories, this study includes Maute andForrester's conceptualization of product information attributesas moderators of the eWOM information–eWOM effectrelationship.

Past studies provide evidence attesting to the notion that thecharacteristics of the product (i.e., the subject of WOMcommunications) may affect how consumers process theWOM message (Sundaram and Webster, 1999).

If consumers have difficulty in judging product quality or ifjudgmental criteria are ambiguous, the value of availableinformation for the purposes of analysis increases (Bone, 1995).This suggests that the influence of WOM may be greater insome situations than in others; specifically, WOM effectsshould be greater when the consumer faces an ambiguoussituation (Bone, 1995).

Heightened perceived risk along the search-experienceproduct continuum, which results from a lack of knowledgeand information, increases the importance of experience (Hsiehet al., 2005). Consumers, especially when they intend topurchase experience goods, can reduce their uncertainty byreferring to eWOM information, such as online consumerreviews, that typically contain expert advice and commentsfrom experienced users. For this reason, the eWOM effect isexpected to be greater for experience goods rather than forsearch goods, leading to the following hypothesis. H3: TheeWOM effect is greater for experience goods than for searchgoods.

2.4. Interaction effect: eWOM direction and product type

Higher levels of brand familiarity, generated through director indirect brand-related experiences, are associated with awell developed knowledge structure about the brand and itsattributes (Alba and Hutchinson, 1987). Given that consumersare not likely to change their attitudes toward familiar brands(Hoyer and MacInnis, 1997), exposure of familiar brand toWOM communications, either positive or negative, is notlikely to produce significant changes in consumers' preexist-ing brand evaluations (Sundaram and Webster, 1999). Lessfamiliar brands, that are usually associated with consumers'underdeveloped knowledge structures, are more susceptibleto change in their brand evaluations following exposure to

Fig. 1. A sample experimental eWOM (positive/search/established).

63C. Park, T.M. Lee / Journal of Business Research 62 (2009) 61–67

any new brand-related communications. Consumers, lessfamiliar with a brand, are more amenable to processing newbrand-related information and changing their brand evalua-tions based on the direction of the information (Sundaram andWebster, 1999). Given that finding search attributes on theInternet is easier than finding experience attributes, con-sumers are likely to possess a much more detailed, richcognitive structure for search goods than for experiencegoods. In addition, Hoch and Ha (1986) suggest thatadvertising's framing (positive/negative) effect is strongerwhen the product is ambiguous, that is, when quality is hardfor a consumer to determine. Therefore, the evaluation ofexperience goods, compared with search goods, may sustaingreater damage in the eWOM effect due to negativeinformation because the negative eWOM information willmagnify consumers' prevailing uncertainty and fear related totheir poor cognitive knowledge structure. These argumentsgive rise to the following hypothesis. H4: The impact differ-ence of negative and positive eWOM information on the eWOMeffect is greater for experience goods than for search goods.

2.5. Interaction effect: website reputation and product type

The message source is a crucial determinant of persuasionespecially when consumers evaluate experience attributes for aproduct (Jain and Posavac, 2001). Jain and Posavac (2001)argue that when consumers have no prior knowledge of abrand's qualities, the use of high-credibility sources commu-nicates experience attribute information more persuasively. Thisstudy includes the website reputation which is expected to serveas extrinsic cues of eWOM information quality especially in thecase of purchasing experience goods.

If the consumer is exposed to eWOM information from ahighly credible source (highly reputable website) endorsingexperience goods, the eWOM information should be morepersuasive than any eWOM information from a less crediblesource. This prediction should hold because it is a main effectprediction: that those highly credible sources (establishedreputation websites) should be more persuasive than sourceswith little or no credibility (unestablished websites). Thisbelievability is even more important in the ambiguouscircumstance of experience goods where consumers do notpossess prior knowledge or experience. H5: The impactdifference of established and unestablished website's reputationon the eWOM effect is greater for experience goods than forsearch goods.

3. Method

3.1. Research design

This study utilized an experiment that manipulated the eWOMinformation direction, the website reputation, and the producttype, thus resulting in a 2 (positive vs. negative)×2 (establishedvs. unestablished)×2 (search vs. experience) between-subjectsfactorial design. Subjects were assigned randomly to the expe-riment's treatments.

3.2. Development of stimuli

An in-depth focus group, involving eight participants, wasconducted to brainstorm for search and experience goods. Fromthe list of 20 products generated, 6 products were then chosen tobe included in the pretest.

Based on Nelson (1974), product type was measured on a 5-point scale (1 = product quality can be evaluated prior topurchase, 5 = product quality cannot be evaluated until purchaseand use of the product) for a sample of 20 males and 20 females.The results indicated that the product type mean for a languageschool program was the highest and the TOIEC (Test of Englishfor International Communication) book was the lowest. Themeans difference of these two products was significant(Mbook=1.5, Mlanguage school =4.4, t(39)=−25.76, p= .000),while evaluations made by male and female respondents forthese two products showed no significant difference. Accordingto the FGI (Focus Group Interview) and pretest results, theTOIEC book for search goods and the language school programfor experience goods were selected for the main experiment.

Development of websites and eWOM. In order to ensureconsistency across treatments, all treatment web pages werecreated from one actual web page, keeping an identical style andlayout throughout, differing only in the Internet shopping mallname, product type, and eWOM information direction, thatconstituted the manipulation.

After reviewing samples of eWOM posted on the website, apanel of doctoral candidates and marketing faculty familiar witheWOM literature generated a list of eight key items for eachproduct.

The website reputation was manipulated by selecting an actualInternet shopping mall. Interpark.com (opened as an Internetshopping mall first in Korea in June, 1996) was selected as anestablished website. Zoamart.com (a start-up Internet shoppingmall in Korea) was selected as an unestablished website.

Fictitious brand names were created (“Point” for the TOEICbook and “Nice” for the language school program) to avoidpossible error sources. Fig. 1 displays sample experimentalstimulus.

Table 1ANCOVA results

Source df MS F-value

CovariateeWOM familiarity 1 6.24 22.98***

eWOM information direction (A) 1 22.54 82.99***Website reputation (B) 1 29.25 107.70***Product type (C) 1 49.45 182.06***A×B 1 0.06 0.20A×C 1 1.31 4.81*B×C 1 2.13 7.86**A×B×C 1 3.75 13.79***Error 431 0.27Total 439

* pb .05, ** pb .01, *** pb .001.

64 C. Park, T.M. Lee / Journal of Business Research 62 (2009) 61–67

3.3. Pretesting

Pretests were conducted with a sample of 20 males and 20females. To assess the positivity/negativity of the framingmanipulation, the subjects rated the extent to which they feltthat the eWOM information stressed the negative/positiveaspects of the discussed object on a 5-point scale (1 =negative information to 5 = positive information). Thewebsite reputation was measured with 10 items on a 5-point semantic differential scale (α=0.92), adapted fromShamdasani et al. (2001), measuring the website's reputationin terms of its expertise and trustworthiness. The order effectwas controlled by randomizing the arrangement of eWOMinformation. The pretest results suggest that the eWOMdirection (Mnegative = 1.5, Mpositive = 4.7, t(39) =−29.24,p= .000) and website reputation (mean value of 10 items)(Munestablished=1.6, Mestablished=4.0, t(39)=28.92, p= .000)measures yield significant differences in the expecteddirection. These results indicate that the manipulations ofthe eWOM information direction and website reputation aresuccessful.

3.4. Measures

The eWOM effect was measured by 3 items on a 5-pointratings of agreement (1 = strongly disagree to 5 = stronglyagree) with the following three statements adapted from Jeonand Park (2003): (1) I will refer to this eWOM information in apurchase decision. (2) Overall, I think this eWOM informationis credible. (3) This eWOM information will crucially affectmy purchase decision. The reliability coefficient for the scalewas 0.73. The eWOM information direction, website reputa-tion, and product type were measured for the same items for thepretests.

Covariates. Past research suggest that brand familiarityaffect consumer purchase decisions (Lane and Jacobson, 1995).Consumers are likely to behave in a similar pattern whenprocessing eWOM information. Thus, eWOM familiarityare likely to affect eWOM-based purchase decisions. Tocontrol for the eWOM familiarity effect, eWOM familiarityis included in the model as a covariate. eWOM familiarity wasmeasured with 3 items on a 5-point semantic differential scale(α=0.82), using statements adapted from Kent and Allen(1994): Regarding eWOM information, I am (1) unfamiliar/familiar, (2) inexperienced/experienced, and (3) not knowl-edgeable/knowledgeable.

3.5. Experimental procedure

Undergraduate students, taking business classes at a majoruniversity in Seoul, had notice of an online experiment. Classassistant requested webmaster to post a notice/link concerningthe online experiment on the several classes' websites. In orderto motivate subjects to respond, the subjects were given freemeal vouchers upon their survey submission. The online surveywas programmed such that respondents would be assignedrandomly to one of the eight conditions in the experiment, with

the cell having the least responses given priority, hence ensuringequal cell sizes. A short description of the websites was given;then, subjects were shown the eWOM information direction,website reputation, and product type combinations as desig-nated by the treatment cell. Following this exposure, eWOMeffects were measured based on the eWOM stimuli the subjectshad just seen. Covariates and manipulation checks forexperimental stimuli were then measured (with the sameitems for pretests), followed by questions about generaldemographics.

4. Analysis and results

A total of 440 responses were collected with 55 responses ineach of the eight treatment conditions. Respondents were aged19–28 with 55.5% men and 44.5% women. All experimentalmanipulations were again assessed and found to be effective inthe main study.

4.1. Hypothesis testing

The hypotheses were tested using a 2×2×2 analysis ofcovariance with eWOM familiarity as covariate. ANCOVAresults are available in Table 1. Table 2 presents the means andstandard deviations of the eight cells involved in the experi-mental design for this study. Fig. 2 illustrates the summary ofhypotheses testing. As shown in Fig. 2, all the hypotheses aresupported. Figs. 3 and 4 display the interaction effects foreWOM information direction-product type and website reputa-tion-product type.

The ANCOVA results (see Table 1) include statisticallysignificant 3-way interaction effect (F(1, 431)=13.79, pb .001).The structure of the observed 3-way interaction appears inFig. 5.

Comparisons show that the impact on the eWOM effectresulting from negative eWOM with established websites issignificantly greater than positive eWOM for search goods (3.6versus 3.0; pb .001) and experience goods (4.4 versus 4.0;pb .001). However, this difference is insignificant in the case ofthe eWOM for search goods posted on the unestablishedwebsites (3.0 versus 2.9; pN .20). In contrast, for experience

Fig. 3. The effects of the eWOM information direction on the eWOM effect as afunction of product type.

Table 2The cell means and standard deviations for the eWOM effect

eWOMinformationdirection

Established reputation Unestablished reputation

Search Experience Search Experience

Positive 3.0 4.0 2.9 3.1 3.3(0.53) (0.44) (0.44) (0.48) (0.63)

Negative 3.6 4.4 3.0 3.9 3.7(0.51) (0.41) (0.68) (0.70) (0.76)

Total 3.3 4.2 3.0 3.5(0.59) (0.47) (0.58) (0.70)

3.7 (0.68) 3.2 (0.70)

Note: The standard deviations are in parentheses.

65C. Park, T.M. Lee / Journal of Business Research 62 (2009) 61–67

goods, negative eWOM with unestablished websites result insignificantly greater impact on the eWOM effect than positiveeWOM (3.9 versus 3.1; pb .001).

5. General discussion, implications, and limitations

5.1. General discussion

First, the eWOM effect is greater for negative eWOM thanfor positive eWOM. In addition, whereas most previousresearch has focused on the negativity effect and the reasonsfor that effect, this research shows that the product typeassociated with the eWOM discussions moderates this effect.Specifically, the negativity effect appears to be more significantwhen the eWOM is for experience goods rather than for searchgoods.

Given that finding search attributes on the Internet is easierthan finding experience attributes, consumers are likely topossess a much more detailed, rich cognitive structure of search

Fig. 2. A summary of h

goods than of experience goods. Empirical results suggest thatexperience goods, compared with search goods, sustain greaterdamage in the eWOM effect due to negative eWOMinformation. The reason is that the negative eWOM informationmagnify consumers' prevailing uncertainty and fear initiated bytheir poor cognitive knowledge structure of experience goods.This research provides an insight into why and how the eWOMnegativity effect is different across product types.

Second, the website reputation has a significant impact onthe eWOM effect. Specifically, the eWOM effect is greater forwebsites with established reputations than for unestablishedwebsites. This result empirically validates the vehicle sourceeffect (website reputation) (Shamdasani et al., 2001) in theeWOM context. This provides an insight into regarding awebsite's reputation as dimensions for source credibility in theeWOM context.

Another interesting finding of this study is that the impact ofa website's reputation on the eWOM effect is greater for

ypotheses testing.

Fig. 4. The effects of website reputation on the eWOM effect as a function ofproduct type.

Fig. 5. The three-way interaction between the eWOM information direction,website reputation, and product type.

66 C. Park, T.M. Lee / Journal of Business Research 62 (2009) 61–67

experience goods than for search goods. According to Zeithaml(1988), consumers depend more on extrinsic cues such asbrand name than intrinsic cues whenever evaluating a product'squality is difficult. The empirical results suggest that awebsite's reputation could serve as extrinsic cues of eWOMinformation quality especially in the case of purchasingexperience goods.

Third, the eWOM effect is greater for experience goods thanfor search goods. In comparison to the offline purchaseenvironment, the online environment is characterized by greateruncertainty because there are limited cues for informationprocessing (Hanson, 2000). In particular, the inability of theonline environment to convey sensory attributes increasesconsumer uncertainty for experience goods. This findingsuggests that the eWOM could serve as a mechanism to reduceuncertainty especially for those who are considering purchase ofexperience goods. The asymmetric impact of the product typeon the eWOM effect suggests that eWOM information is moreuseful for experience goods than for search goods.

5.2. Implications

This research provides a number of academic and managerialimplications. First, the empirical results of this study show thatnegative eWOM has greater purchase influence than positiveeWOM. This negative eWOM effect is especially augmentedfor the case of experience goods. This study makes a theoreticalcontribution to marketing communication literature by empiri-cally validating the moderating role of product type in theeWOM information direction-eWOM effect relationship. Thisstudy also has managerial implications for marketing commu-nication managers. The empirical results show that negativeeWOM for experience goods should remain a high priority ifonly selective monitoring is possible because of marketingbudget constraints. This study makes a managerial contributionto marketing practitioners by suggesting priority guidelines onthe monitoring of eWOM information direction and producttype.

Second, Internet marketers may offset perceived risk byenhancing positive eWOM on well-established sites. Thisresearch shows that this strategy could be more effective forexperience goods, the quality of which is inherently difficult forconsumers to judge, than for search goods. This implies that a

well-chosen website for eWOM placement can build andenhance eWOM effectiveness.

Finally, this study illustrates that the eWOM effect is greaterfor experience goods than for search goods. Therefore, Internetmarketers intending to utilize eWOM strategically should makeevery effort to strengthen the perceived usefulness of onlinereview especially for experience goods. For example, MISSHA,a small-sized manufacturer of cosmetics in Korea, hassuccessfully transformed experience goods such as cosmeticsinto search goods by encouraging eWOM regarding consumers'positive product experience. The manufacturer provides variouskinds of incentives (e.g. points) for useful online reviews.

5.3. Limitations and further research

This research report suffers from several limitations. First,this research focuses on product attribute-related eWOMinformation. However, previous literature suggests that theperceived value of negative information for analytical purposesis likely to be higher in the morality (e.g., company values)versus the ability (e.g., product attributes) domain, which leadsto the greater weighting of negative information in the moralitydomain (Skowronski and Carlston, 1987). Thus, the findings ofthis study may not be generally applicable to negative eWOMinformation reflecting a company's values (e.g., its neglect ofsocial responsibility). Further research could examine this issuein an experimental setting by manipulating the type of negativeinformation (values-related versus attributes-related) (Ahluwa-lia et al., 2000). Second, this study does not include consumercharacteristics (e.g. involvement and commitment toward thebrand or product) in the experiment. Previous research suggeststhat the commitment of the consumer toward the brand is amoderator of negative information effects (Ahluwalia et al.,2000). Specifically, the negative effect appears to be more likelywhen a consumer's commitment to the target object or issue islow. Subsequent research could address this issue in an eWOMcontext. Finally, future research could investigate an eWOMeffect model on cross-cultural basis. Considering that eWOM isglobal, cross-cultural research on the eWOM effect would be aninteresting issue.

67C. Park, T.M. Lee / Journal of Business Research 62 (2009) 61–67

References

Ahluwalia Rohini, Burnkrant Robert E, Unnava H Rao. Consumer response tonegative publicity: the moderating role of commitment. J Mark Res 2000;37(May):97–108.

Alba Joseph W, Hutchinson J Wesley. Dimensions of consumer expertise.J Consum Res 1987;13(March):411–54.

Arndt J. Role of product-related conversations in the diffusion of a new product.J Mark Res 1967;4(August):291–5.

Bone Paula Fitzgerald. Word-of-mouth effects on short-term and long-termproduct judgments. J Bus Res 1995;32(3):213–23.

Brown Jacqueline Johnson, Reingen Peter H. Social ties and word-of-mouthreferral behavior. J Consum Res 1987;14(December):350–62.

Chatterjee P. Online reviews: do consumers use them? Adv Consum Res2001;28:129–33.

Dholakia Ruby Roy, Sternthal Brian. Highly credible sources: persuasivefacilitators or persuasive liabilities? J Consum Res 1977;3(March):223–32.

Gruen TW, Osmonbekov T, Czaplewski AJ. eWOM: the impact of customer-to-customer online know-how exchange on customer value and loyalty. J BusRes 2006;59(4):449–56.

Hanson Ward. Principles of Internet Marketing. Cincinnati, OH: South-WesternCollege Publishing; 2000.

Hennig-Thurau Thorsten, Gwinner Kevin P, Walsh Gianfranco, GremlerDwayne D. Electronic word-of-mouth via consumer-option platforms:what motives consumers to articulate themselves on the Internet? J InteractMarket 2004;18(1):38–52.

Hoch Stephen J, Ha Young-Won. Consumer learning: advertising and theambiguity of product experience. J Consum Res 1986;13(October):221–33.

Homer PM, Yoon S. Message framing and the interrelationships among ad-based feelings, affect, and cognition. J Advert 1992;21(1):19–33.

Hoyer WD, MacInnis Deborah J. Consumer Behavior. MA: Houghton MufflinCo; 1997.

Hsieh Yi-Ching, Chiu Hung-Chang, Chiang Mei-Yi. Maintaining a committedonline customer: a study across search-experience-credence products.J Retail 2005;81(1):75–82.

Jain Shailendra Pratap, Posavac Steven S. Prepurchase attribute verifiability,source credibility and persuasion. J Consum Psychol 2001;11(3):169–80.

Jeon SY, Park HJ. The influence of information characteristics on word-of-mouth effect. J Consum Stud 2003;14:21–44.

Kent RJ, Allen CT. Competitive interference effects in consumer memory foradvertising: the role of brand familiarity. J Mark 1994;58(July):97–105.

Klein Lisa R. Evaluating the potential of interactive media through a new lens:search versus experience goods. J Bus Res 1998;41(3):195–203.

Lane Vicki, Jacobson Robert. Stock market reactions to brand extensionannouncements: the effects of brand attitude and familiarity. J Mark 1995;59(January):63–77.

Maute Manfred F, Forrester William R. The effect of attribute qualities onconsumer decision making: a causal model of external information search.J Econ Psychol 1991;12:643–66.

Mizerski RW. An attribution of the dispropotionate influence of unfavorableinformation. J Consum Res 1982;9(December):301–10.

Nelson P. Advertising as information. J Polit Econ 1974;82(4):729–54.Richins ML. Negative word-of-mouth by dissatisfied consumers: a pilot study.

J Mark 1983;47(Winter):68–78.Shamdasani Prem N, Stanaland Andrea JS, Tan Juliana. Location, location,

location: insights for advertising placement on the Web. J Advert Res2001;41(4):7–21.

Skowronski JJ, Carlston DE. Social judgment and social memory: the role ofcue diagnosticity in negativity, positivity, and extremity biases. J Pers SocPsychol 1987;52(4):689–99.

Skowronski JJ, Carlston DE. Negativity and extremity in impression formation:a review of explanations. Psychol Bull 1989;105(January):131–42.

Sundaram DS, Webster C. The role of brand familiarity on the impact of word-of-mouth communication on brand evaluations. Adv Consum Res1999;26:664–70.

Weinberger Marc G, Allen Chris T, Dillon William R. Negative information:perspectives and research directions. In: Monroe Kent, editor. Advances inConsumer Research. Ann Arbor, MI: Association for Consumer Research;1981. p. 398–404.

Wright P. The harassed decision maker: time pressures, distractions, and the useof evidence. J. Appl. Psychol. 1974;59(5):555–61.

Zeithaml Valarie A. Consumer perceptions of price, quality, and value: a means-end model and synthesis of evidence. J Mark 1988;52(3):2–22.